wildmatch.c

   1 /*
   2 **  Do shell-style pattern matching for ?, \, [], and * characters.
   3 **  It is 8bit clean.
   4 **
   5 **  Written by Rich $alz, mirror!rs, Wed Nov 26 19:03:17 EST 1986.
   6 **  Rich $alz is now <rsalz@bbn.com>.
   7 **
   8 **  Modified by Wayne Davison to special-case '/' matching, to make '**'
   9 **  work differently than '*', and to fix the character-class code.
  10 */
  11
  12 #include "cache.h"
  13 #include "wildmatch.h"
  14
  15 typedef unsigned char uchar;
  16
  17 /* What character marks an inverted character class? */
  18 #define NEGATE_CLASS    '!'
  19 #define NEGATE_CLASS2   '^'
  20
  21 #define FALSE 0
  22 #define TRUE 1
  23
  24 #define CC_EQ(class, len, litmatch) ((len) == sizeof (litmatch)-1 \
  25                                     && *(class) == *(litmatch) \
  26                                     && strncmp((char*)class, litmatch, len) == 0)
  27
  28 #if defined STDC_HEADERS || !defined isascii
  29 # define ISASCII(c) 1
  30 #else
  31 # define ISASCII(c) isascii(c)
  32 #endif
  33
  34 #ifdef isblank
  35 # define ISBLANK(c) (ISASCII(c) && isblank(c))
  36 #else
  37 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
  38 #endif
  39
  40 #ifdef isgraph
  41 # define ISGRAPH(c) (ISASCII(c) && isgraph(c))
  42 #else
  43 # define ISGRAPH(c) (ISASCII(c) && isprint(c) && !isspace(c))
  44 #endif
  45
  46 #define ISPRINT(c) (ISASCII(c) && isprint(c))
  47 #define ISDIGIT(c) (ISASCII(c) && isdigit(c))
  48 #define ISALNUM(c) (ISASCII(c) && isalnum(c))
  49 #define ISALPHA(c) (ISASCII(c) && isalpha(c))
  50 #define ISCNTRL(c) (ISASCII(c) && iscntrl(c))
  51 #define ISLOWER(c) (ISASCII(c) && islower(c))
  52 #define ISPUNCT(c) (ISASCII(c) && ispunct(c))
  53 #define ISSPACE(c) (ISASCII(c) && isspace(c))
  54 #define ISUPPER(c) (ISASCII(c) && isupper(c))
  55 #define ISXDIGIT(c) (ISASCII(c) && isxdigit(c))
  56
  57 /* Match pattern "p" against "text" */
  58 static int dowild(const uchar *p, const uchar *text, int force_lower_case)
  59 {
  60         uchar p_ch;
  61
  62         for ( ; (p_ch = *p) != '\0'; text++, p++) {
  63                 int matched, special;
  64                 uchar t_ch, prev_ch;
  65                 if ((t_ch = *text) == '\0' && p_ch != '*')
  66                         return ABORT_ALL;
  67                 if (force_lower_case && ISUPPER(t_ch))
  68                         t_ch = tolower(t_ch);
  69                 if (force_lower_case && ISUPPER(p_ch))
  70                         p_ch = tolower(p_ch);
  71                 switch (p_ch) {
  72                 case '\\':
  73                         /* Literal match with following character.  Note that the test
  74                          * in "default" handles the p[1] == '\0' failure case. */
  75                         p_ch = *++p;
  76                         /* FALLTHROUGH */
  77                 default:
  78                         if (t_ch != p_ch)
  79                                 return NOMATCH;
  80                         continue;
  81                 case '?':
  82                         /* Match anything but '/'. */
  83                         if (t_ch == '/')
  84                                 return NOMATCH;
  85                         continue;
  86                 case '*':
  87                         if (*++p == '*') {
  88                                 const uchar *prev_p = p - 2;
  89                                 while (*++p == '*') {}
  90                                 if ((prev_p == text || *prev_p == '/') ||
  91                                     (*p == '\0' || *p == '/' ||
  92                                      (p[0] == '\\' && p[1] == '/'))) {
  93                                         /*
  94                                          * Assuming we already match 'foo/' and are at
  95                                          * <star star slash>, just assume it matches
  96                                          * nothing and go ahead match the rest of the
  97                                          * pattern with the remaining string. This
  98                                          * helps make foo/<*><*>/bar (<> because
  99                                          * otherwise it breaks C comment syntax) match
 100                                          * both foo/bar and foo/a/bar.
 101                                          */
 102                                         if (p[0] == '/' &&
 103                                             dowild(p + 1, text, force_lower_case) == MATCH)
 104                                                 return MATCH;
 105                                         special = TRUE;
 106                                 } else
 107                                         return ABORT_MALFORMED;
 108                         } else
 109                                 special = FALSE;
 110                         if (*p == '\0') {
 111                                 /* Trailing "**" matches everything.  Trailing "*" matches
 112                                  * only if there are no more slash characters. */
 113                                 if (!special) {
 114                                         if (strchr((char*)text, '/') != NULL)
 115                                                 return NOMATCH;
 116                                 }
 117                                 return MATCH;
 118                         }
 119                         while (1) {
 120                                 if (t_ch == '\0')
 121                                         break;
 122                                 if ((matched = dowild(p, text,  force_lower_case)) != NOMATCH) {
 123                                         if (!special || matched != ABORT_TO_STARSTAR)
 124                                                 return matched;
 125                                 } else if (!special && t_ch == '/')
 126                                         return ABORT_TO_STARSTAR;
 127                                 t_ch = *++text;
 128                         }
 129                         return ABORT_ALL;
 130                 case '[':
 131                         p_ch = *++p;
 132 #ifdef NEGATE_CLASS2
 133                         if (p_ch == NEGATE_CLASS2)
 134                                 p_ch = NEGATE_CLASS;
 135 #endif
 136                         /* Assign literal TRUE/FALSE because of "matched" comparison. */
 137                         special = p_ch == NEGATE_CLASS? TRUE : FALSE;
 138                         if (special) {
 139                                 /* Inverted character class. */
 140                                 p_ch = *++p;
 141                         }
 142                         prev_ch = 0;
 143                         matched = FALSE;
 144                         do {
 145                                 if (!p_ch)
 146                                         return ABORT_ALL;
 147                                 if (p_ch == '\\') {
 148                                         p_ch = *++p;
 149                                         if (!p_ch)
 150                                                 return ABORT_ALL;
 151                                         if (t_ch == p_ch)
 152                                                 matched = TRUE;
 153                                 } else if (p_ch == '-' && prev_ch && p[1] && p[1] != ']') {
 154                                         p_ch = *++p;
 155                                         if (p_ch == '\\') {
 156                                                 p_ch = *++p;
 157                                                 if (!p_ch)
 158                                                         return ABORT_ALL;
 159                                         }
 160                                         if (t_ch <= p_ch && t_ch >= prev_ch)
 161                                                 matched = TRUE;
 162                                         p_ch = 0; /* This makes "prev_ch" get set to 0. */
 163                                 } else if (p_ch == '[' && p[1] == ':') {
 164                                         const uchar *s;
 165                                         int i;
 166                                         for (s = p += 2; (p_ch = *p) && p_ch != ']'; p++) {} /*SHARED ITERATOR*/
 167                                         if (!p_ch)
 168                                                 return ABORT_ALL;
 169                                         i = p - s - 1;
 170                                         if (i < 0 || p[-1] != ':') {
 171                                                 /* Didn't find ":]", so treat like a normal set. */
 172                                                 p = s - 2;
 173                                                 p_ch = '[';
 174                                                 if (t_ch == p_ch)
 175                                                         matched = TRUE;
 176                                                 continue;
 177                                         }
 178                                         if (CC_EQ(s,i, "alnum")) {
 179                                                 if (ISALNUM(t_ch))
 180                                                         matched = TRUE;
 181                                         } else if (CC_EQ(s,i, "alpha")) {
 182                                                 if (ISALPHA(t_ch))
 183                                                         matched = TRUE;
 184                                         } else if (CC_EQ(s,i, "blank")) {
 185                                                 if (ISBLANK(t_ch))
 186                                                         matched = TRUE;
 187                                         } else if (CC_EQ(s,i, "cntrl")) {
 188                                                 if (ISCNTRL(t_ch))
 189                                                         matched = TRUE;
 190                                         } else if (CC_EQ(s,i, "digit")) {
 191                                                 if (ISDIGIT(t_ch))
 192                                                         matched = TRUE;
 193                                         } else if (CC_EQ(s,i, "graph")) {
 194                                                 if (ISGRAPH(t_ch))
 195                                                         matched = TRUE;
 196                                         } else if (CC_EQ(s,i, "lower")) {
 197                                                 if (ISLOWER(t_ch))
 198                                                         matched = TRUE;
 199                                         } else if (CC_EQ(s,i, "print")) {
 200                                                 if (ISPRINT(t_ch))
 201                                                         matched = TRUE;
 202                                         } else if (CC_EQ(s,i, "punct")) {
 203                                                 if (ISPUNCT(t_ch))
 204                                                         matched = TRUE;
 205                                         } else if (CC_EQ(s,i, "space")) {
 206                                                 if (ISSPACE(t_ch))
 207                                                         matched = TRUE;
 208                                         } else if (CC_EQ(s,i, "upper")) {
 209                                                 if (ISUPPER(t_ch))
 210                                                         matched = TRUE;
 211                                         } else if (CC_EQ(s,i, "xdigit")) {
 212                                                 if (ISXDIGIT(t_ch))
 213                                                         matched = TRUE;
 214                                         } else /* malformed [:class:] string */
 215                                                 return ABORT_ALL;
 216                                         p_ch = 0; /* This makes "prev_ch" get set to 0. */
 217                                 } else if (t_ch == p_ch)
 218                                         matched = TRUE;
 219                         } while (prev_ch = p_ch, (p_ch = *++p) != ']');
 220                         if (matched == special || t_ch == '/')
 221                                 return NOMATCH;
 222                         continue;
 223                 }
 224         }
 225
 226         return *text ? NOMATCH : MATCH;
 227 }
 228
 229 /* Match the "pattern" against the "text" string. */
 230 int wildmatch(const char *pattern, const char *text, int flags)
 231 {
 232         return dowild((const uchar*)pattern, (const uchar*)text,
 233                       flags & FNM_CASEFOLD ? 1 :0);
 234 }